Print version ISSN 0042-9686
Bull World Health Organ vol.84 n.7 Genebra Jul. 2006
Mise en place des conditions nécessaires à un suivi favorable à l'équité des progrès en direction des Objectifs du Millénaire pour le développement relatifs à la santé maternelle et infantile
Creación de un marco para que la monitorización de los Objetivos de Desarrollo del Milenio relacionados con la salud materna e infantil sea sensible a la equidad
Meg E WirthI; Deborah BalkII,1; Enrique DelamonicaIII; Adam StoreygardII; Emma SacksII; Alberto MinujinIV
IIndependent consultant to Task Force 4, Child Health and Maternal Health, UN Millennium Project at the time this research was undertaken
IICenter for International Earth Science Information Network (CIESIN), Columbia University; Lamont-Doherty Earth Observatory, 61 Route 9W, PO Box 1000, Palisades, NY 10964, USA
IIIGlobal Policy Section, Division of Policy and Planning, United Nations Children's Fund (UNICEF)
IVGraduate Program on International Affairs (GPIA), The New School for Social Research, New York, USA. Global Policy Section, Division of Policy and Planning, UNICEF at the time this research was undertaken
OBJECTIVE: This analysis seeks to set the stage for equity-sensitive monitoring of the health-related Millennium Development Goals (MDGs).
METHODS: We use data from international household-level surveys (Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS)) to demonstrate that establishing an equity baseline is necessary and feasible, even in low-income and data-poor countries. We assess data from six countries using 11 health indicators and six social stratifiers. Simple bivariate stratification is complemented by simultaneous stratification to expose the compound effect of multiple forms of vulnerability.
FINDINGS: The data reveal that inequities are complex and interactive: inferences cannot be drawn about the nature or extent of inequities in health outcomes from a single stratifier or indicator.
CONCLUSION: The MDGs and other development initiatives must become more comprehensive and explicit in their analysis and tracking of inequities. The design of policies to narrow health gaps must take into account country-specific inequities.
OBJECTIF: La présente analyse s'attache à définir les conditions nécessaires à un suivi favorable à l'équité des progrès en direction des Objectifs du Millénaire pour le développement (OMD) relatifsà la santé maternelle et infantile.
MÉTHODES: Des données tirées d'enquêtes internationales auprès des ménages (Enquêtes démographiques et de santé, DHS) et d'enquêtes en grappes à indicateurs multiples (MICS) ont servi à démontrer la nécessité et la faisabilité de l'établissement d'un point de comparaison pour l'équité, même pour les pays à faible revenu et pour lesquels les éléments disponibles sont rares. Les données provenant de six pays ont été soumises à une évaluation utilisant onze indicateurs sanitaires et considérant six couches sociales. En plus de la stratification bivariée simple, il a été procédé à une stratification simultanée, destinée à mettre en évidence l'effet multifactoriel des formes multiples de vulnérabilité.
RÉSULTATS: Les données font apparaître la complexité des inéquités et l'existence d'interactions entre elles : il est impossible de formuler des déductions quant à la nature et à l'ampleur de ces inéquités à partir d'un paramètre de stratification ou d'un indicateur unique.
CONCLUSION: Les OMD et autres initiatives en faveur du développement devront intégrer une analyse et un suivi plus complets et plus explicites des inéquités. Dans la conception des politiques visant à réduire les disparités en matière d'accès à la santé, il convient de prendre en compte les inéquités spécifiquesà chaque pays.
OBJETIVO: Este análisis trata de crear el marco para lograr que la monitorización de los Objetivos de Desarrollo del Milenio (ODM) relacionados con la salud sea sensible a la equidad.
MÉTODOS: Hemos utilizado los datos de encuestas domiciliarias internacionales (encuestas sobre demografía y salud, y encuestas de conglomerados con múltiples indicadores) para demostrar que es necesario y posible efectuar una descripción basal de las inequidades, incluso en los países de bajos ingresos que disponen de escasos datos. Hemos analizado en seis países 11 indicadores sanitarios estratificados en función de seis variables sociales. La estratificación bivariada simple se complementó con la estratificación simultánea para poner de manifiesto el efecto compuesto de múltiples formas de vulnerabilidad.
RESULTADOS: Los datos revelan que las inequidades son complejas e interactivas y que no es posible hacer inferencias acerca de la naturaleza o la magnitud de las inequidades de los resultados sanitarios a partir de un único indicador o factor estratificador.
CONCLUSIÓN: Tanto en los ODM como en otras iniciativas de desarrollo debe realizarse un análisis y seguimiento más integral y explícito de las inequidades. La formulación de políticas para reducir las diferencias en materia de salud deben tener en cuenta las inequidades específicas de cada país.
Inequities in health are pervasive within countries, rich and poor alike. Even in countries where aggregate health indicators are improving, some health gaps between population groups are widening or remaining stagnant. The size and dynamics of these gaps vary considerably, depending not only on the indicator and country studied, but also the means of stratifying the population into social groups. And yet, health equity analyses too often remain simplistic or nonexistent, even in key development initiatives like the Millennium Development Goals (MDGs) and Poverty Reduction Strategy Papers (PRSPs).
From an ethical and human rights perspective, narrowing avoidable disparities in health is imperative.1 An explicit and systematic commitment to equity must be made to ensure that poor, marginalized and vulnerable groups are given access to health services and opportunities for healthy lives.2,3
Many recent studies have focused either on single health outcomes or on one or two stratifiers: results have demonstrated that inequities in health outcomes differ between and within countries and confirmed the conventional wisdom that ill-health is more prevalent in poor populations than in better-off groups.415 Other research has shown the extent to which spending on health and social services disproportionately favours privileged groups,16,17 quantifying the differences between populations with respect to access to health care and health outcomes. Other work has drawn attention to the wider set of social determinants that stratify health.1820
Here, we use population-based surveys to analyse several indicators and stratifiers, and aim to show that equity analyses in country-level adaptations of the MDGs and PRSPs should be more comprehensive. We also aim to show that: multiple health indicators give a more complete picture of inequalities in health; social disadvantage must be examined holistically to reflect its complexity beyond wealth; the measurement of inequalities is feasible with use of current data even in very poor countries; and the health MDGs should be framed in equity-sensitive terms.
We used data from recent Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS). Countries that lack vital registration systems currently rely on information from population-based surveys to monitor progress towards MDGs. This approach is generalizable to most resource-poor countries that have at least one population-based household survey per country containing information on health and social characteristics. We examine data from six countries across 11 health indicators and six social stratifiers.
Countries were selected to correspond with UN Millennium Project case studies. Data sources used were the DHS surveys for Cambodia (2000), the Dominican Republic (2002), Ethiopia (2000), Ghana (1998), and Kenya (1998), and the Tajikistan 2000 MICS.2126 The Tajikistan data come from aggregate tables distributed by UNICEF. All measures are calculated using DHS data at the individual level or derived from the DHS reports and web site. Where possible, indicator definitions were harmonized across the five DHS countries. Some indicators reported differ from those in DHS reports:27 for example, values of "don't know" or "missing" were excluded from our analysis, whereas in DHS reports these categories are sometimes explicitly reported, or considered equivalent to "no". Similarly, DHS report contraceptive prevalences for women currently in union, whereas we report rates for all women. We report mean age at marriage instead of median age.
We recoded ethnicities into dominant, not dominant, and secondary dominant categories to create larger classes of stratifiers.28,29 A "wealth by poverty line" variable was created with use of existing wealth indices30,31 to complement the stratification by wealth quintile with a simple policy-relevant distinction between just two groups: "poor" and "not poor". Data on the percentage of population living below the poverty line were applied to the wealth index data to create this variable.32,33
We selected health indicators that would match the MDG child health and maternal health indicators, with a few exceptions (Table 1). The nature of the indicators varies, ranging from outcomes (underweight, child mortality), to access to care or preventative interventions (skilled attendant at birth, measles and diphtheriapertussistetanus (DPT) vaccination and contraceptive prevalence), to knowledge (about acquired immunodeficiency syndrome (AIDS)), to fertility-related or women's status indicators, such as age at first marriage.
An equity analysis requires division of a population into groups according to underlying social advantage. The social stratifier most frequently associated with inequity is wealth measured by the set of assets the family has, rather than by monetary income or expenditure. However, stratification by wealth alone is not the most appropriate way to measure inequities in health; in countries with extreme poverty, the wealthiest quintile often resides only in the capital. Furthermore, measurements of wealth at a household level do not capture intra-household inequalities, such as those conferred by gender, age or position within the household family structure.
Multiple dimensions of inequality exist within countries such as age, residence (urban or rural), gender, ethnicity, occupation, geographic survey region, and education level. The health gaps between these groups may be as significant as the gaps between rich and poor. Choice of stratifiers (and health measures) for official monitoring purposes must be based on health and human rights challenges and policy needs and opportunities in each country.34,35 Here we use six key stratifiers to illustrate our overarching point about the need for more nuanced equity analysis: sex, education status, urban or rural residence, ethnicity, wealth, and geographic region of residence (Table 2). The full dataset with 20 indicators for six countries is available elsewhere.36
Our selection of variables is not exhaustive and is constrained by the availability of data in the study countries. However, the stratifiers we have chosen may serve as proxies for other factors of interest. For example, the education stratifier is an imperfect proxy of women's empowerment.
The number of regions per survey varies with the size of the sample and other factors. Especially when used in combination with another stratifier, sample sizes in individual regions can become too small to yield meaningful results. Examination of interaction effects between stratifiers allows for the quantification of cumulative disadvantages of multiple risks. Thus, although simultaneous stratification is important, we note that when the sample sizes are low results should be interpreted with caution.
Cross-classification of indicators captures the complexity of health disadvantage. We did simple stratification (bivariate analysis) for 11 health indicators and wherever possible, we calculated the values for health indicators for all stratifiers (Table 3). To assess their effect both independently and interactively, simultaneous stratification (trivariate analysis) was then performed for pairs of stratifiers. For example, ethnic group health-outcomes were classified by sex, region, residence, wealth, etc., to determine the compounded effect of variables. Some pairings were not generated in the simultaneously stratified analysis because the resulting subgroup was too small or non-existent (e.g. ethnicity with regions). Likewise, mortality indicators were not included in the simultaneous stratification, because the number of events (deaths) was too small to construct robust rates.
Multivariate analysis was not undertaken so as to preserve a simple study design and ensure that the methods could be easily replicated. Finally, we assessed statistical significance of the inequities in health status to identify where gaps result from random variation rather than the statistically valid considerations sought for evidence-based policy-making.
We did a between-means comparison for every stratification class (e.g. education) to test the null hypothesis that there is no statistically significant difference between values of an indicator for all classes defined by the stratifc fier (e.g. none, primary, secondary or more). Similar tests were carried out for selected portions of the simultaneously stratified data. We interpret differences where P <0.05. Tests of significance were not performed on the mortality rate indicators, because they are rates rather than proportions. National-level standard errors from DHS reports can be used as a general indication of likely significance between groups for national mortality rates.
Previous published work suggests that most indicators are differentiated by wealth quintile, with less differentiation where interventions tend towards being universal. In general, we expected rural health outcomes to be worse than those in urban areas, that poor people would have worse outcomes than those categorized as not poor and we expected a certain degree of heterogeneity between regions and across ethnic groups. We expected that education status of the mother would have an important effect on all health indicators.
In Cambodia, Ethiopia, Ghana, and Kenya, there was a significant correlation between education, ethnicity, region and residence and underweight. In Ethiopia and Kenya underweight was also significantly related to wealth and whether a child lives above or below the poverty line. Ethnicity and region and not wealth were found to have the widest range of values for underweight in Ghana. In Ethiopia, the pattern is slightly different with region and education of the mother showing the widest range of values ethnicity seems to be less important here. In Kenya, the pattern differs again, with maternal education, ethnicity, region and wealth quintile all showing roughly equivalent ranges of values. In the simultaneous stratification for Kenya, for women with primary or with secondary or more education levels, the proportion of underweight children is 24 times as great for the children in the poorest households compared with those in the wealthiest households. Rural children are more likely to be underweight, especially in families where the mother has no education or only primary education.
Somewhat unexpectedly, in Ethiopia, wealth does not seem to prevent children from being underweight. Even in the highest wealth quintile, education is a more important factor: children of mothers with no education are twice as likely to be underweight and six times as likely to be severely underweight. In Cambodia, the urban bias is concentrated in mothers who completed schooling (P = 0.0228 for primary education, P = 0.0173 for secondary education, but P = 0.8210 for no schooling). Among those with no formal education in Cambodia, there is no difference between rural and urban levels of underweight children. There was no significant correlation between sex and underweight status in any country studied.
For most countries studied, diphtheriapertussistetanus (DPT) and measles immunization are significantly stratified by not only wealth quintile but also by maternal education status, ethnicity and region. Urban versus rural residence also stratified all immunization indicators for Ethiopia and Ghana, and this disparity improves significantly with maternal education status in Cambodia. Immunization rates in Tajikistan varied between regions from just above 60% to over 90%.
Surprisingly, for all countries studied, sex was not a significant factor in immunization rates at the bivariate level, with the exception of DPT3 in the Dominican Republic. Rural versus urban residence was not a strong factor in immunization disparities in Cambodia and Kenya. In Tajikistan, there is no stepwise correlation between wealth and immunization status. In Kenya, simultaneous stratification shows that ethnicity is correlated with immunization, with less dominant ethnic groups falling well behind dominant groups. Boys and girls were immunized at about the same rates for measles, but DPT3 rates differed by sex, especially in the non-dominant ethnic group (Table 4). Simultaneous stratification for Ethiopia also reveals gender inequity in measles immunization: daughters of women with secondary or more schooling have higher rates than do sons (P = 0.03) (Table 5). Sex-based differences are also evident in Kenya, with 98% of urban boys vaccinated against measles compared with 90% of urban girls (P = 0.07). Thus in several countries it seems that basic immunization is inequitably distributed, suggesting significant challenges for implementation of vertical programmes.
Child mortality rates
In Cambodia, Ethiopia, Ghana and Kenya educational level of the mother, region and residence stratify under-five mortality rates (U5MR). In Kenya, ethnicity dramatically stratifies U5MR, with a range 31253 across groups. Additionally, the expected stepwise decrease in U5MR with increasing wealth quintile is observed. The capital of Cambodia, Phnom Penh, consistently shows the lowest mortality, with mortality rates in the next best region almost twice as high. In Ghana, inequality in childhood mortality is closely aligned with differences in education and place of residence: more highly educated women and urban dwellers have much lower child mortality. And in Ethiopia, educational level of the mother significantly stratifies mortality in neonates, infants and children under 5 years.
By contrast, in Ethiopia, wealth quintile and urban/rural distinctions are not particularly strong stratifiers of health outcomes. In fact, the richest quintile differs little from the poorest. In Kenya, it seems that the difference between no maternal education and primary education does not yield large disparities in U5MR. Likewise in Ghana, primary education actually yields a higher neo-natal and infant mortality than does no education.
Usage of skilled birth attendants
Maternal education, ethnicity, region, residence and wealth quintile are all significantly correlated with usage of skilled birth attendants in Ethiopia, Ghana and Kenya. For instance, in Ethiopia, major differences are evident when the indicator is stratified by educational level with only 3% of those with no education using skilled birth attendants, 10% of those with primary education and 45% of women with secondary education or more. In Kenya, the Mijikenda/Swahili ethnic groups were at a low of 27% usage and the Kikuyu at a high of 71%. Likewise, in Ghana, ethnicity seems to have an important effect on use of skilled birth attendants, with a near twofold, statistically significant difference between the primary dominant (63%) and the not dominant groups (34%) (P <0.00005). The non-poor are almost twice as likely as the poor to have a skilled birth attendant in Kenya. In Cambodia, almost 90% of the births in Phnom Penh are assisted by skilled attendants, in stark contrast with a national average of only one-third. Education and rural/urban residence also stratify skilled birth attendant use in Cambodia (P <0.00005). In Tajikistan, 55% of the lowest wealth quintile and 87% of the highest quintile use skilled birth attendants, and the rural/urban differential is 68% versus 84%.
In Kenya, simultaneous stratification reveals dramatic inequities by maternal education, region and urban versus rural residence even among the non-poor (Table 6). In the Dominican Republic, where national rates of skilled birth attendant coverage are relatively high, there exists relative equity with respect to maternal education and urban versus rural residence.
In Ethiopia, Ghana and Kenya, both indicators for AIDS knowledge are stratified significantly by maternal education, ethnicity, region and residence, suggesting a rather unequal spread and uptake of critical information and education about HIV/AIDS. In Cambodia, knowledge that a healthy-looking person may have AIDS and that using a condom during sex can help prevent HIV infection is significantly stratified by maternal education status, despite high overall knowledge (the national average is above 80% for both indicators). In Tajikistan, rural populations have much lower levels of AIDS knowledge. Wealth differentiates only the richest group 20% of the top quintile know that condoms help prevent infection, compared with less than 5% for the rest of the population and large differences exist between regions. In the Dominican Republic, knowledge varies by region, with a range 7896% for the indicator on "a healthy-looking person may have AIDS".
There was strong regional variation in most of the countries studied and ethnic variation was especially pronounced in Ethiopia and Ghana.
In Ethiopia, Ghana and Kenya, contraceptive prevalence (using a modern method) is correlated significantly by all stratifiers. In Tajikistan, there is a clear educational gradient, with those with no education at 16%, 26% for secondary education and 41% for tertiary. By wealth quintile, there is relatively more equitable distribution of contraceptive use, although the richer groups have greater prevalence.
Surprisingly, in the Dominican Republic the percentage of women using a modern method of contraception declines as education increases, and the differences are statistically significant. Among women with no education, the rate of contraceptive use is significantly higher in urban areas (P = 0.0143), but in women with primary education, use rates are significantly higher in rural areas (P = 0.0299). Contraceptive use decreases significantly with education at all levels in urban areas, and from primary to secondary in rural areas. Region and residence are the main stratifiers in Cambodia, with no significant correlation between formal education and contraceptive use. In Ethiopia, the expected education effect applied only in the capital.
Age at first marriage
For all of the countries with information for this indicator, the data revealed statistically significant educational gradients those with secondary education married at least a year and in some cases 4 years later than did those with no education. Rural women married earlier than did urban women. In most countries, regional variations were observed this was particularly striking, with nearly a 5-year difference found, amongst regions of Ethiopia.
Ethnicity was not found to be a significant stratifier in most countries, Ethiopia being one exception with large differentials similar to those found by region. Wealth quintile had a greater effect on age at first marriage in Kenya (2.6 years' difference from lowest to highest quintile) than did education. In Cambodia, the difference between urban and rural dwellers is not statistically significant in those with no education or those with secondary education but it is larger and significant for those with primary education.
Inequities in health exist even in the poorest countries. Our results highlight the wide variation that different indicators and social stratifiers exhibit within countries. This variation coupled with the robust findings of disparity suggest that the MDGs should be monitored in an equity-sensitive manner, starting with a baseline description of inequities across a range of health indicators. Population-based surveys can be used to establish such an equity baseline even in data-poor countries. Tracking of progress in reducing disparities should complement overall monitoring of the health MDGs.
Several limitations of our analysis deserve mention here. Despite the richness of the data, this brief snapshot of health inequalities is not intended to form the complete baseline in the countries considered. Data sources other than DHS and MICS may be more appropriate to track all health indicators in a manner explicitly tailored to national circumstances. Subsampling from the vital registration system, demographic surveillance system (DSS) data and facility-based surveys are important complementary sources of infc formation when available.37,38 Shortcomings in sampling frames often result in vulnerable groups such as refugee populatc tions, urban slum dwellers, orphans and linguistic minorities being excluded from survey analyses.
Our results confirm that the currente focus on pro-poor health policies is an oversimplification that omits other core sources of health inequities.39 Stratification by wealth, ethnicity, maternal education status, sex, region and urban/rural residence yielded statistically significant differences across a wide range of health indicators in six countries. In many cases, the ethnic, educational and regional variations were more pronounced than were the disparities attributable to differences in wealth. Furthermore, analysis of dual forms of marginalization reveals the complexity of health gaps within countries.
The region of residence stratifier is often coterminous with those of ethnic divisions or poverty profiles, although this association is only revealed by simultaneous stratification. For example, measles vaccination rates seem to vary considerably by wealth, but when regions are added as substrata it becomes clear that some districts represent the bottom quintiles of the population. While wealth is an important focus, the geographic elements of poverty would have been overlooked without disaggregation of the data. An understanding of the correlates of poverty will be an important element in reducing deprivation. The results of our analysis suggest that in many countries, reducing inequality in health will require policies to be tailored by geographic area. Thus, geographic identifiers should be added to all surveys, including MICS and DHS, to allow countries to georeference survey information.
Educational status of mothers is a critical social determinant of most health indicators. Investments in education must be seen as having a dual positive effect in both the education and health sectors. Simultaneously, health messages and programmes should be designed to reach mothers with low education status and their children. And ethnicity, a core form of marginalization, remains under-studied in the health and development literature.
Importantly, different health indicators yielded different patterns of inequity. For example, AIDS knowledge may be high and only somewhat evenly distributed between groups, but rates of delivery by a skilled birth attendant and U5MR within the same country may be grossly inequitable (as in Cambodia). Inferences about the nature or extent of inequities in health cannot be drawn from a single indicator. Nor can we assume that groups disadvantaged in one indicator are necessarily disadvantaged in another. Our analysis strongly suggests that reliance on single indicators alone and certainly national averages would lead to limited, misguided recommendations for policy.
Countries should start with a clear health (in)equity baseline based on the MDGs but tailored to their unique sociocultural dynamics. Once the (in)equity baseline has been established, the difficult work begins. What are the policies and programmes that will address these critical issues? Standard behavioural and social science methods must also be used to explain and augment the data and analysis described here. Multivariate quantitative analysis and qualitative studies are required to clarify causal pathways that lead certain groups to be disadvantaged relative to others.
And the health MDGs indeed all relevant MDGs must be reframed to prioritize marginalized groups. Equitable progress towards the MDG targets would mean that the health outcomes of the disadvantaged improve at the same or faster rates as the better-off groups.2,39 Poverty reduction strategies, a key instrument of current development policy, must be synchronized with the MDGs.40 Then, policy changes aligned with PRSP and MDG priorities ought to be designed and tracked so as to measure progress from the (in)equity baseline.
Health exclusion results from multiple and overlapping forms of social exclusion, in addition to differences in health systems. The full array of underlying social determinants of health must be addressed in both health research and development policy.19 And rather than a patchwork of "pro-poor" interventions and ad hoc targeted programmes, universal health systems dedicated to the inclusion of all population groups are needed to build more efficient, equitable and healthier societies. Analysis of the type presented here is a feasible first step towards these goals and towards equitable achievement of the MDGs.
This article is an abbreviated version of a Background Paper prepared for Task Force 4 (Child Health and Maternal Health) of the UN Millennium Project. We are grateful to Paula Braveman for input at all stages of the background paper, Mushtaque Chowdhury and Cesar Victora for input during the early stage of conceptualization of the paper, Davidson Gwatkin and Jeanette Vega for comments on a near final draft and Lynn Freedman of Task Force 4 for support throughout the process. We also thank James Connolly for statistical and editorial support and Meredith Slopen for her help with editing the final submission.
Funding: the World Bank's Japan Policy and Human Resource Development (PHRD) Fund provided funding under a grant to Columbia University's Earth Institute and CIESIN (http://ciesin.columbia.edu/povmap/).
Competing interests: none declared.
1. Braveman P, Gruskin S. Poverty, equity, human rights and health. Bull World Health Organ 2003; 81:539-45. [ Links ]
2. Freedman L, Waldman R, de Pinho H, Wirth M, Chowdhury AMR, Rosenfield A for the UN Millennium Project Task Force on Child Health and Maternal Health. Who's got the power? Transforming health systems for women and children. London: Earthscan; 2005. [ Links ]
3. Gwatkin DR. Who would gain most from efforts to reach the Millennium Development Goals for health? An inquiry into the possibility of progress that fails to reach the poor. HNP Discussion Paper. Washington, DC: World Bank; 2002. [ Links ]
4. Gwatkin DR, Rutstein S, Johnson K, Suliman EA, Wagastaff A. Initial country-level information about socio-economic differences in health, nutrition, and population. Washington, DC: World Bank; 2003. [ Links ]
5. Minujin A, Delamonica E. Mind the gap! Widening child mortality disparities. J Hum Dev 2003,4:397-418. [ Links ]
6. Victora CG, Wagstaff A, Schellenberg JA, Gwatkin DR, Claeson M, Habicht JP. Applying an equity lens to child health and mortality: more of the same is not enough. Lancet, 2003:362:233-41. [ Links ]
7. Segone M. Child poverty in Niger: longitudinal, gender, spatial and wealth analysis based on DHS 1992, MICS 1996, DHS 1998 and MICS 2000. Niamey, Niger: UNICEF; October 2001. [ Links ]
8. Evans T, Whitehead M, Diderichsen F, Bhuiya A, Wirth M. Challenging inequities in health: from ethics to action. New York: Oxford University Press; 2001. [ Links ]
9. AbouZahr C, Wardlaw T. Antenatal care in developing countries: promises, achievements and missed opportunities: An analysis of trends, levels and differentials, 1990-2001. Geneva: WHO/UNICEF; 2004. [ Links ]
10. Victora CG, Vaughan, JP, Barros FC, Silva AC, Tomasi E. Explaining trends in inequities: evidence from Brazilian child health studies. Lancet 2000;356:1093-8. [ Links ]
11. Wagstaff A. Poverty and health sector inequalities. Bull World Health Organ 2003;80:97-105. [ Links ]
12. Kunst A, Houweling T. A global picture of poor-rich differences in the utilisation of delivery care. In: Van Lerberghe W, De Brouwere V (editors). Safe motherhood strategies: a review of the evidence. Antwerp: ITG Press; 2001. [ Links ]
13. Schellenberg JA, Adam T, Mshinda H, Masanja H, Kabadi G, Mukasa O, et al. Inequities among the very poor: healthcare for children in rural southern Tanzania. Lancet 2003;361:561-6. [ Links ]
14. Wagstaff A,Watanabe N. Socioeconomic inequalities in child malnutrition in the developing world. Policy Research Working Papers. Washington DC: World Bank; 2000. [ Links ]
15. Wagstaff A, Bustreo F, Bryce J, Claeson M, Axelsson H. Inequalities in child health: are we narrowing the gap? HNP Discussion Paper. Washington, DC: World Bank; 2003. [ Links ]
16. Castro-Leal F, Dayton J, Demery L, Mehra K. Public spending on health care in Africa: Do the poor benefit? Bull World Health Organ 2000;78:66-74. [ Links ]
17. McIntyre D Gilson L in collaboration with EQUINET, the Regional Network for Equity in Health in Southern Africa. Equitable health care financing and poverty challenges in the African context. Paper Presented to Forum 9, Global Forum for Health Research. Mumbai, September 12-16, 2005. Available from http://www.equinetafrica.org/bibl/docs/McIfin092005.pdf [ Links ]
18. Casas JA, Dachs NW, Bambas A. Health disparities in Latin America and the Caribbean: the role of social and economic determinants. In: Equity and health: views from the Pan American sanitary bureau. Washington, DC: Pan American Health Organization; 2001. [ Links ]
19. Marmot M. Social determinants of health inequalities. Lancet 2005; 365:1099-104. [ Links ]
20. Vandemoortele J. Absorbing social shocks, protecting children and reducing poverty: the role of basic social services. UNICEF staff working papers, Evaluation, Policy and Planning Series no. 00-001. New York: UNICEF; 2000. [ Links ]
21. National Institute of Statistics, Directorate General for Health [Cambodia], and ORC Macro. Cambodia Demographic and Health Survey 2000. Phnom Penh and Calverton (MD), USA: National Institute of Statistics, Directorate General for Health, and ORC Macro; 2001. [ Links ]
22. National Council for Population and Development (NCPD), Central Bureau of Statistics (CBS) (Office of the Vice President and Ministry of Planning and National Development) [Kenya], and Macro International Inc. (MI). Kenya Demographic and Health Survey 1998. Calverton (MD): NDPD, CBS, and MI; 1999. [ Links ]
23. Ghana Statistical Service (GSS) and Macro International Inc. (MI). Ghana Demographic and Health Survey 1998. Calverton (MD): GSS and MI; 1999. [ Links ]
24. Central Statistical Authority [Ethiopia] and ORC Macro. Ethiopia Demographic and Health Survey 2000. Addis Ababa and Calverton (MD): Central Statistical Authority and ORC Macro; 2001. [ Links ]
25. Centro de Estudios Sociales y Demográficos (CESDEM) and MEASURE DHS+/ORC Macro. República Dominicana Encuesta de Demografía y Salud 2002. [Dominican Republic Demographic and Health Survey 2002]. Calverton, MD: Centro de Estudios Sociales y Demográficos, CESDEM and MEASURE DHS+/ORC Macro; 2003. [In Spanish.] [ Links ]
26. United Nations Children's Fund (UNICEF). Multiple Indicator Cluster Survey (MICS): Tajikistan. Dushanbe: UNICEF; 2000. [ Links ]
27. Rutstein SO, Rojas G. Guide to DHS Statistics. Calverton (MD): ORC Macro; 2003. [ Links ]
28. Brockerhoff M, Hewett P. Ethnicity and child mortality in sub-Saharan Africa. Policy Research Division Working Paper No. 107. New York: Population Council; 1998. [ Links ]
29. Moyo S. Socio-economic dominance of ethnic and racial groups the African experience. Human Development Report Office. Occasional Paper, Background Paper for HDR 2004. 2004/8. New York: United Nations Development Programme; 2004. [ Links ]
30. Filmer D, Pritchett L. Estimating wealth effects without expenditure data or tears: an application to educational enrollments in states of India. Demography 2001;38:115-32. [ Links ]
31. Sahn DE, Stifel DC. Poverty comparisons over time and across countries in Africa. World Development 2000;28:2123-55. [ Links ]
32. UNDP (United Nations Development Programme). Human Development Report 2001: making technology work for human development. New York: Oxford University Press; 2001. Available from: http://hdr.undp.org/reports/global/2001/en/. [ Links ]
33. UNDP (United Nations Development Programme). Human Development Report 2003: Millennium Development Goals: a compact among nations to end poverty. New York: Oxford University Press; 2003. [ Links ]
34. Bambas Nolen L, Braveman P, Dachs JNW, Delgado I, Gakidou E, Moser K, et al. Strengthening health information systems to address health equity challenges. Bull World Health Organ 2004;83:597-603. [ Links ]
35. Braveman P. Monitoring inequities in health: a policy-oriented approach in low- and middle-income countries. Geneva: World Health Organization; 1998. WHO document WHO/CHS/HSS/98.1. [ Links ]
36. Wirth M, Balk D, Storeygard A, Sacks E, Delamonica E, Minujin A. Three way tables: setting the stage for equity-sensitive monitoring of the health Millennium Development Goals. Available from: http://www.ciesin.columbia.edu/pdf/Two_Way_Health_Equity_Tables.pdf [ Links ]
37. Ngom P, Phillips J, Pence B, MacLeod B. Demographic surveillance and health equity in sub-Saharan Africa: past and present efforts. African Population and Health Research Center. Working Paper Series. Nairobi: APHRC; 2003. [ Links ]
38. Vega J, Irwin A. Tackling health inequalities: new approaches in public policy. Bull World Health Organ 2004;82:482. [ Links ]
39. Minujin A, Delamonica E. Equality matters for a world fit for children: lessons from the 90s. UNICEF Staff Working Papers, Division of Policy and Planning Series (Draft, Number DPP-03), New York: UNICEF; 2003. [ Links ]
40. UN Millennium Project. Investing in development: a practical plan to achieve the Millennium Development Goals. Overview. New York: Millennium Project; 2005. [ Links ]
(Submitted: 1 December 2004 Final revised version received: 3 December 2005 Accepted: 14 December 2005)